Analysis of the influencing factors and spatial heterogeneity characteristics of women’s safety perceptions from street-view images: taking Tianhe District of Guangzhou city as an example

The utilization of streets has proven to have noticeable differences among social groups. However, the actual management of such street usage has generally ignored gender differences, especially the sensitive and vulnerable needs of women, weakening the sense of security and inclusiveness of urban street scenes. In this study, the Baidu map street-view images were used to score the safety perception of a wide range of street scenes within Tianhe District, Guangzhou City, using human-machine adversarial machine deep learning with big data from street-view images combined with MGWR analysis. Focusing on analyzing the visual environmental influences and spatial heterogeneity of women’s sense of security, this study identifies the streets that require improvement for the sense of security in the city. It clarifies the environmental and spatial elements that need to be improved in the streets. This study focuses on conducting a current situation analysis and experience summary of the safety perception of streets in Tianhe District from a female perspective. It is conducive to proposing precise and effective street safety design strategies for street spaces, providing investigation, research methods, and optimization approaches for improving street safety perception in other regions or cities, contributing to safer and more inclusive urban governance.